For many years, the 'gene' has been defined as a length of DNA which encodes a single protein. Findings in the last few years, however, have made molecular biologists realize that the conventional definition of a gene is inadequate.
The discovery of alternative splicing, whereby the same exons (expressing regions) are brought together in different combinations to encode different proteins, made researchers realize that genes are more complex than they had thought, but did not neccesitate a radical change in the concept of the gene.
Now, it turns out that the storage of information by DNA is far more complex that was previously thought. Recent work has shown that the genome is full of overlapping transcripts, with some coding regions running straight through into adjacent ones. Some transcripts are even contained entirely within other coding regions, and it is now clear that both strands of DNA (that is, the 'sense' and 'antisense' strands) can encode proteins.
Whereas genes were believed to occupy discrete regions of chromosomes, it appears that the exons of some genes can be spread out over hundreds of thousands of kilobases. Messenger RNAs can even be churned out from different chromosomes and spliced to produce protein-coding transcripts.
The role of RNA in epigenetic (or 'extragenomic') mechanisms has been known for some time, but it now seems clear that the molecule once regarded as the humble messenger has a greater role in transacting and passing information from one generation to the next.
When the sequencing of the human genome was completed, many were surprised to discover that it contained only 30,000 genes. That number, though, was based on the conventional 'one gene, one protein' definition of a gene.
Karen Eilbeck, a co-ordinator of the Sequence Ontology Consortium, says that it took 25 scientists nearly two days to reach a definition of a gene that they were all happy with: "a locatable region of genomic sequence, corresponding to a unit of inheritance, which is associated with regulatory regions, transcribed regions and/or other functional sequence regions."
Mo Costandi 
Not very satisfying is it. What is a “unit of inheritance”? Presumably they mean a region small enough to not be broken apart by recombination easily..well that doesn’t make sense. And what do they mean by “associated with”: just correlation or a causal relationship?
Guess that’s what happens when you define something by committee.